Explainable AI-Based Phishing Website Detection Using Hybrid Machine Learning for Enhanced Cybersecurity

Abstract

The expeditious growth of online services has crucially increased the number and sophistication of phishing attacks, making phishing websites one of the most serious cybersecurity threats to individuals and organizations. Conventional machine learning-based phishing detection methods often achieve high prediction exactness but function as black-box models, offering limited insight into the reasoning behind their decisions. This lack of transparency reduces user confidence and hinders adoption in security-critical applications. To mark these challenges, this paper proposes an Explainable AI-Based Phishing Website Detection Framework Using Hybrid Machine Learning that combines the prognostic capabilities of multiple machine learning algorithms with interpretable decision-making. The raised framework extracts comprehensive URL, domain, webpage, and security-related features, followed by data refinement, feature selection, and hybrid ensemble classification to accurately discern phishing websites from legitimate ones. To enhance model transparency, Explainable Artificial Intelligence (XAI) techniques such as SHAP (SHapley Additive exPlanations) are embedded to identify and visualize the influence of each feature toward the final prediction. This enables security analysts and end users to understand why a website is classified as phishing, thereby improving trust and supporting informed decision-making. The proposed framework is investigated using benchmark phishing datasets and compared with standard machine learning approaches based on accuracy, precision, recall, F1-score, ROC-AUC, and explainability. Experimental results are expected to demonstrate that the proposed hybrid model not only achieves superior detection reliability but also provides meaningful and transparent explanations, making it a reliable solution for next-generation intelligent phishing website detection systems.

Keywords